Table 8 Results of the robustness test.

From: Effects of digital trade export on wage inequality among enterprises: evidence from China

 

Model

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Variable

Maximum likelihood estimation

Two-step consistent estimation

Digital trade at the city level

Internet intensity with two- period lag

Internet intensity with three- period lag

Use alibb instead of treat

Quadratic model

treat × lInternetest × lInternetmt

−0.0027***

−0.0031***

–

–

–

–

−0.0038***

(−3.7683)

(−5.3543)

(−4.1835)

treat × lDTst

–

–

−0.0111***

–

–

–

–

(−7.3208)

treat × l2Internetest × l2Internetmt

–

–

–

−0.0045***

–

–

–

(−3.3670)

treat × l3Internetest × l3Internetmt

–

–

–

–

−0.0030*

–

–

(−1.7246)

alibb × lInternetest × lInternetmt

–

–

–

–

–

−0.0025***

–

(−5.6199)

(treat × lInternetest × lInternetmt)2

–

–

–

–

–

–

0.0001

(0.7357)

Constant

−2.7483***

−2.3833***

−0.9166***

14.2403

52.9962***

−1.1641***

−1.0894***

(−920.3544)

(−488.0579)

(−14.6354)

(0.8381)

(2.5851)

(−25.9462)

(−11.2432)

Other control variables

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Sample size

367,861

367,861

346,042

360,689

360,232

957,752

367,861

R2_a

–

–

0.3984

0.4965

0.5096

0.5176

0.4966

chi2

–

262,233.99

–

–

–

–

–

chi2_c

77,381.51

12,607.34

–

–

–

–

–

  1. The numbers in parentheses refer to t statistics; ***, **, and * are p values at 1%, 5%, and 10%, respectively; R2_a refers to the adjusted R2. The chi2 and the chi2_c represent Wald chi2 statistic value and Wald chi2 statistic value of lnsigma2 in the heteroscedasticity test, respectively. Different regression models select different interaction terms for their core explanatory variables, thus controlling the corresponding single variables for each interaction term.
Back to article page